[BioC] F-tests for factorial effects - limma

Naomi Altman naomi at stat.psu.edu
Tue Dec 21 23:21:19 CET 2004


I am analyzing a 2-factor factorial Affy experiment, with 3 d.f. for each 
factor.

I would like to get the F-tests for the main effects and interactions using 
limma.

I have computed all the contrasts, and got the t-tests (both unadjusted and 
eBayes).  I do know how to combine these into F-tests "by hand" but I could 
not figure out if there was a simple way to do this using limma.

I had a look at FStat (classifyTestsF).  There seems to be a problem, in 
that the matrix tstat is not premultiplied by the contrast matrix when the 
F-statistics are computed.  So, if the contrasts are not full-rank, an 
error is generated (instead of the F-statistics) because nrow(Q) != 
ncol(tstat)..  (See the lines below).

if (fstat.only) {
         fstat <- drop((tstat%*% Q)^2 %*% array(1, c(r, 1)))
         attr(fstat, "df1") <- r
         attr(fstat, "df2") <- df
         return(fstat)
     }

I figured that before I fiddled with the code, I would check to make sure 
that I didn't miss an existing routine to do this.

Thanks in advance.

Naomi S. Altman                                814-865-3791 (voice)
Associate Professor
Bioinformatics Consulting Center
Dept. of Statistics                              814-863-7114 (fax)
Penn State University                         814-865-1348 (Statistics)
University Park, PA 16802-2111



More information about the Bioconductor mailing list